RSTP topology change. Issue:port in Listening state for 30 seconds?

[Karlis Tucs]

Hi. I am trying to solve problem of our customer regarding long re-convergence times for Cisco Business 350 Series Managed Switches.

In attachment I have added diagram of switches and console messages in all 3  states of topology. I will shortly describe issue below, but it should be clear from images alone also.

3 Switches connected in loop. One port is declared alternate port. I remove 1 connection and this alternate port must become designated port in order to send and receive data. It does this, BUT before it goes to forwarding state it stays in Listening(?) state for 30 seconds. I have read RSTP standard and other sources, and I can't seem to understand why it should take 30 seconds for RSTP.  I thought it would be the case for STP. Our customer loses data for 30 seconds and it is unacceptable from his point of view.

Can someone please englighten me and help me solve this issue?

[M02@rt37]

Hello @Karlis Tucs 

The reason why it takes 30 seconds for the alternate port to transition from Listening to Forwarding state in RSTP is due to the way the protocol operates. When a topology change occurs, RSTP must first wait for the max_age timer (by default, 20 seconds) to expire before it can start the Topology Change Notification [TCN] process. Once the TCN is generated, it takes up to 10 seconds to propagate throughout the network, and then each switch starts its own TC While timer, which lasts for 15 seconds.

During this TC While timer*, each switch listens for other switches to send TC messages, indicating that they have updated their forwarding table. Once the TC While timer expires, the switch can then transition the alternate port to forwarding state.

Therefore, the total time it takes for the alternate port to transition to forwarding state in RSTP after a topology change is up to 20 + 10 + 15 = 45 seconds.

To reduce the downtime caused by a topology change, you can:

-- Configure RSTP with a lower max_age timer value. This will cause the TCN process to start sooner, reducing the overall downtime.

-- Use a redundant link to provide alternate connectivity to the network, so that if a link fails, traffic can be immediately switched to the alternate link.

-- Implement fast link failure detection mechanisms, such as LACP, to quickly detect and recover from link failures.

* The TC While timer is a timer used in the RSTP to control the length of time that a switch listens for topology change information from other switches in the network.

[MHM Cisco World]

why you make your design like this 
config PO between Access Sw and Core SW 
this will protect you from link failure and the only case the STP re-elect again is when the Core SW failed 

[Karlis Tucs]

This is typical design seen in many places where switches are connected in loop.

What does PO mean?

Also, if I use CBS350-8T-E-2G instead of CBS350-16P-2G switch model it takes less than 1 second :] . So there is something fishy here, I am still trying to figure out what theoretically should be the time for     Discarding Alternate port -> Discarding Designated port-> Forwarding Designated port.